Land use mapping methodology using remote sensing for the regional planning directives in Segovia, Spain

This study presents the methodology followed in the land use mapping at scale 1:50,000 for the Functional Area of Segovia, Spain. The study is included in the regional planning directives for Segovia, and makes a comprehensive use of the remote sensing techniques developed during the second half of the 1990s. The methodological precedents in this respect are analyzed, mainly the CORINE land cover project, together with the data sources and the techniques that were used to obtain the thematic information searched for. The less known methods are described and a complete methodological sequence of the work is offered. The main characteristics of the methodology are the high degree of automation, objectivity, possibility of direct contrasting and its capacity for quick updating. Diverse data sources have been used, such as cartographic vector information and satellite imagery. LANDSAT-TM and IRS-1D Pan were used as well as aerial oblique photography. All information was integrated into a Geographic Information System (GIs; named SIGIM-TD). Data fusion methods were also widely used to improve the spatial resolution of the images. A neural network was generated to provide an appropriate classification method. Results of the neural network-based classification are shown and a classification-fieldwork correlation of 0.887 was obtained, in contrast with coefficients of 0.334, 0.432, 0.234 and 0.678 achieved through other techniques. Graphic results of the work are presented together with the data sources used for the map elaboration. Finally, a discussion on the advances that these techniques represent is done. (C) 2002 Elsevier Science B.V. All rights reserved.